Tamarixia radiata global distribution to current and future climate using the climate change experiment (CLIMEX) model

The phloem-limited bacteria, “Candidatus Liberibacter asiaticus” and “Ca. L. americanus”, are the causal pathogens responsible for Huanglongbing (HLB). The Asian citrus psyllid Diaphorina citri Kuwayama (Hemiptera: Liviidae) is the principal vector of these “Ca. Liberibacter” species. Though Tamarixia radiata Waterston (Hymenoptera: Eulophidae) has been useful in biological control programmes against D. citri, information on its global distribution remains vague. Using the Climate Change Experiment (CLIMEX) model, the potential global distribution of T. radiata under the 2050s, 2070s, and 2090s for Special Report on Emissions Scenarios A1B and A2 was defined globally. The results showed that habitat suitability for T. radiata covered Africa, Asia, Europe, Oceania, and the Americas. The model predicted climate suitable areas for T. radiata beyond its presently known native and non-native areas. The new locations predicted to have habitat suitability for T. radiata included parts of Europe and Oceania. Under the different climate change scenarios, the model predicted contraction of high habitat suitability (EI > 30) for T. radiata from the 2050s to the 2090s. Nevertheless, the distribution maps created using the CLIMEX model may be helpful in the search for and release of T. radiata in new regions.

www.nature.com/scientificreports/ and 9:00 am relative humidity, and 15:00 h 69 . The Fifth Assessment Report (AR5) published by the Intergovernmental Panel on Climate Change-IPCC presents four updated greenhouse gas trajectories (Representative Concentration Pathways-RCPs) to replace the SRES scenarios. Compared to the current Report scenarios (AR5), the A2 SRES scenario is equivalent to RCP 8.5, as it presents similar forecasts until the end of the century. The A2 SRES scenario predicts an increase in atmospheric concentrations of CO 2 by 846 ppm and an increase in temperature of 6 °C at the end of 2100, while its RCP 8.5 equivalent indicates an increase of 7 °C in Temperature and CO 2 concentrations of 936 ppm 70 . Associated with this, the A2 SRES scenario incorporates representative data on technology, demographics, and economic variables related to greenhouse gases (GHG) from independent and self-sufficient countries, which gives it proven consistency in its assumptions 67 .
Parameters used in CLIMEX. Moisture parameters. In CLIMEX, the moisture content is established by four parameters, being the lower limit of soil moisture (SM0), the optimal lower soil moisture (SM1), the optimal moisture of the upper soil (SM2), and the upper limit of soil moisture (SM3) 71 . We determined the lower soil moisture threshold (SM0) and the upper soil moisture threshold (SM3) from the best fit of the model in the software and according to the global distribution of T. radiata. Also, we used relative humidity to define the lower optimum soil moisture (SM1) and upper optimum soil moisture (SM2), the parametric value provided by the temperate model in CLIMEX and the parametric value provided by the temperate model in CLIMEX 64 . The lower (SM0), ideal (SM1 and SM2), and upper (SM3) limits established were 0.07, 0.8, 1, and 3 (Table 1), respectively.
Temperature parameters. In CLIMEX, temperatures are defined in four parameters, the lower temperature limit (DV0), the lower optimum temperature (DV1), the upper optimum temperature (DV2), and the upper temperature limit (DV3) 60 . Variables DV1 and DV2 represent the most favourable temperature range for the species. The temperature requirements of T. radiata have already been reported, so the lower temperature limit (DV0) used in the model was 15 °C because the insect does not emerge below this temperature 55,63 . As for the lower optimum temperature (DV1) and the upper optimum temperature (DV2), they were set at 20ºC and 30ºC, respectively, these temperatures being ideal for the growth and establishment of T. radiata 55,62,65,72 . The uppertemperature limit (DV3) was 35 °C, which has low insect parasitism rates 47,55,63 .
Stress parameters. Stresses are characterized by non-ideal environmental conditions that restrict the establishment of a species in a region 71 . In CLIMEX, four types of stress parameters are defined, namely: CS (cold stress), HS (heat stress), DS (drought stress), and WS (moisture stress) 73 . The stress parameters used in our models were cold stress degree day threshold (DTCS), cold stress degree day rate (DHCS), heat stress temperature threshold (TTHS), temperature rate stress threshold (THHS), dry stress threshold (SMDS), dry stress ratio (HDS), wet stress threshold (SMWS) and wet stress ratio (HWS). The values for the stress parameters were established according to the best fit in the software according to the regions of occurrence of T. radiata and in the parametric value provided by the Mediterranean and temperate model in CLIMEX (Table 1). www.nature.com/scientificreports/ Cold stress. The development of insects can be influenced by temperature, as they are ectothermic organisms 74,75 . Low temperatures can affect the development of T. radiata, in which there is no oviposition below 10 °C 72 . Therefore, the degree day threshold (DTCS) was set at 10 °C and the stress accumulation rate (DHCS) was set at -0.001 to adjust the insect distribution in the occurrence areas.
Heat stress. Oviposition and development of T. radiata are not possible at temperatures above 35 °C 72,76 . Moreover, T. radiata exposed to heat treatment (38 °C) for 15 min survived 77 . However, when the heat stress was maintained for 2 h, about 65% of T. radiata died. Thus, we considered the 37 °C temperature threshold (TTHS) to be the best fit of the model outputs to the areas of T. radiata and stress accumulation rate (THHS) at 0.00001 week −1 (Table 1).
Dry stress. Considering the regions of occurrence of T. radiata, the dry stress threshold (SMDS) was adjusted to 0.1, and the dry stress accumulation rate (HDS) was fixed at − 0.01 week −1 (Table 1) covering temperate regions.
Wet stress. The parameters of wet stress (SMWS) and stress to the accumulation rate (HWS) were defined based on the CLIMEX parameters for humid tropical regions, which are similar to the insect's distribution regions and the best fit of the output of the insect model. Therefore, SMWS was set to 2.5, and HWS was set to 0.1 week −1 ( Table 1).

Model validation.
We evaluated the CLIMEX model performance based on the distribution of the species, mainly in the regions of America and Asia, where higher occurrences were observed. The verification demonstrates reliability in the final model, and most distribution data are inserted in areas with a high Ecoclimatic Index (Fig. 3).
Human or animal rights. This article does not contain any studies with human participants or animals performed by any of the authors.

Results
Model validation. The distribution of the species, particularly in the parts of America and Asia where higher occurrences were noted, was used to validate the model. In addition, the habitat suitability for T. radiata obtained from the model settings in Table 1 covered both native and non-native present distribution points of the parasitoid. This verification demonstrates reliability in the model's predictions, and most of the distribution data were found in areas with high Ecoclimatic Index (Fig. 3).  www.nature.com/scientificreports/ Potential global distribution of T. radiata. Under the current time, the model predicts that suitable areas for the establishment of T. radiata are found in the world's tropics and subtropical climates (Fig. 4). The predicted suitable areas exceeded the known distribution points of the parasitoid with high habitat suitability (for EI > 30) covering all continents except Antarctica. The areas with high suitability for T. radiata occur in parts of Brazil, Mexico, and the USA in the Americas; Ghana, Nigeria, Kenya, and South Africa in Africa; China and India; and Australia and Papua New Guinea in Oceania.
In the future scenario (SRES A1B), the potential global distribution of T. radiata shows a contraction in areas that were projected to be optimal in the current climate (Fig. 5). Specifically, the model predicts that low suitability (0 < EI < 30) will increase, while high suitability (EI > 30) for the parasitoid will decrease from the 2050s to 2090s. The model predicts that by 2050, areas in the Americas, Africa, Asia, and Oceania will all be suitable for T. radiata. These include parts of Uruguay, Paraguay, Argentina, Brazil, Nicaragua, Cayenne, Guyana, Venezuela, Peru, Colombia, and Honduras. The areas that will continue to have high suitability for T. radiata include parts of Brazil, Paraguay, Uruguay, Argentina, and Nicaragua in the Americas; Tanzania, Uganda, Madagascar, Cameroon, and South Africa in Africa; and China and Indonesia in Asia.
Under the SRES A2 scenario, the results showed that areas highly suitable for the parasitoid would be concentrated mainly in parts of Brazil, Surname, Uruguay, Paraguay, Peru, Argentina, Colombia and the USA in the Americas; Madagascar, Tanzania, South Africa and Kenya in Africa; China and Indonesia in Asia; and Papua New Guinea in Oceania (Fig. 6). The prediction shows contraction of suitable areas from the current time until the 2090s. In the future, areas with high habitat suitability for T. radiata mainly occur in countries, such as Paraguay, Uruguay, Brazil, and Argentina in the Americas; Madagascar, Kenya and Tanzania in Africa; China and Indonesia in Asia; Australia and Papua New Guinea in Oceania; and Italy, Spain, Portugal and Greece in Europe.

Discussion
Natural enemies, such as parasites, predators and parasitoids, are sensitive to temperature changes and may be affected by climate through extrinsic and intrinsic mechanisms 78 . Consequently, global warming is expected to induce a shift in the ecological range of many species, thereby causing habitats that are presently suitable to become unsuitable for their establishment in the future 5 . If pests migrate into areas where their natural enemies are absent, the ability of these biological control agents to keep them in check will reduce. However, a new natural enemy community may help provide some level of control 79 . As the earth warms, natural enemies of herbivores, in particular, may find it difficult to parasitize on their host effectively 80 . Moreover, changes in temperature, humidity, and soil moisture patterns, as influenced by climate change, may have substantial implications on the population and behaviour of natural enemies because farmers are likely to use adaptive management practices to adjust to climate change 79 .
In this study, the CLIMEX model was used to define the potential global geographical distribution of T. radiata, using the physiological stress factors of the parasitoid. Our predictive results were consistent with the historical distribution records of T. radiata. The model's prediction was reliable as assessed by predictive performance in its native and non-native areas. We found that the majority (61.49%) of these historical records fell www.nature.com/scientificreports/ within the areas predicted to be highly suitable for the parasitoid, followed by 34.63% in areas with low suitability, and then 3.88% of the points occurring in areas of unsuitability. In its native range, low to high EI values of habitat suitability for T. radiata were found in most parts of Asia, where it is believed to be the aboriginal home of the parasitoid. 11,23 The areas predicted to be suitable for T. radiata, were also predicted to have suitability for D. citri. 5 Despite biotic and abiotic factors considered in the present study, our model predicts that habitat suitability for T. radiata could expand outside its presently known native and non-native areas. Specifically, parts of the world that showed expansion of the suitable regions but have not recorded T. radiata, include Kenya, Tanzania, Ethiopia, Uganda, and Nigeria in Africa; Australia and Papua New Guinea in Oceania; Thailand and Cambodia in Asia; and Portugal and Spain in Europe. Moreover, our model predicts that large areas in Africa are suitable for the parasitoid, such as Nigeria, Kenya, Nigeria and Tanzania where D. citri is present 5,23,27 . Thus, researchers can utilize our maps to create ecologically acceptable management plans against D. citri in continents where it is present, such as Asia and the Americas 12 .
The CLIMEX model shows that the potential distribution of T. radiata is primarily centered within tropical and subtropical climates, with a few habitat suitability in the Mediterranean climates. This habitat suitability www.nature.com/scientificreports/ for T. radiata is likely to be widely distributed within tropical climates, with habitat suitability ranging from low to high, probably due to its warm temperatures throughout the year 80 . Within the subtropical climates, areas below the equator showed either low or unsuitability for T. radiata. In contrast, the most suitable climate areas within the subtropical climates above the equator ranged from low to high habitat suitability for the T. radiata. The predictions show that the highly suitable areas in Australia are confined to a narrow margin along the eastern and western coasts, with most of the inland areas, south and northern parts of the country having unsuitability to low habitat suitability for T. radiata. According to earlier reports 47,54 , the establishment of T. radiata is likely to occur in areas with warm and dry climates, where temperatures do not exceed the lower and upper thresholds of 12 and 35 °C, respectively, for the development and survival of life stages 47,54,63,65 . Moreover, transcriptome analysis of T. radiata showed that heat stress significantly induced the transcription of immunological response, stress signaling transduction, and oxidation resistance, including highly expressed heat shock proteins, ATPases, and detoxifying enzymes 77 . Ramos Aguila et al. 35 , found that T. radiata's host-feeding activity is temperature-dependent and varied across temperature regimes: the host-feeding rate increased as the temperature increased up to 30 °C, started decreasing after this temperature, and decreased to its lowest level at 35 °C. www.nature.com/scientificreports/ When T. radiata was exposed to different temperature regimes, the highest levels of fecundity, net reproduction rate, intrinsic growth rate, and maximum growth rate were observed at 27.5 °C, and population growth was faster at temperatures ranging between 27.5 and 30 °C 36 .
In the USA, our modelling results show that T. radiata is distributed more narrowly in the country, primarily along the southern coast of the states (i.e., North and South Carolina, Mississippi, Louisiana, New Mexico, Arizona and California). Furthermore, the model predicts that entire states, such as Florida and Texas, are suitable for T. radiata. For instance, in Texas, favourable winter weather conditions are warm and dry with occasional frosty nights, followed by suitable summer conditions that are hot and humid, and moderately hot. During summer, the minimum, and maximum temperatures in Florida range from 32 to 35 °C, although mean summer temperatures are above 21 °C in other states across the southern parts of the USA (Florida Automated Weather Network at https:// fawn. ifas. ufl. edu).
Under CSIRO-Mk3.0 GCM for the SRES A1B and A2, the model predicts that the suitable global areas for T. radiata will decrease from the 2050s to the 2090s. However, some areas, like the northern fringes of Africa, will become more suitable for T. radiata in the future. This suggests that future climate change will alter the geographic distribution of T. radiata depending on the geographical region. Moreover, global warming will cause some countries within subtropical climates, such as Greece, Italy, and Portugal, to have a marginal expansion of suitable habitats for T. radiata. This supports previous studies which demonstrate that climate change will affect the geographical distribution of many species in the future [81][82][83][84] .
Notwithstanding the validity and reliability of our model predictions, we need to mention that certain limitations or drawbacks should be considered when interpreting any species distribution models. In this study, our CLIMEX employed climate-related factors, meteorological datasets and distribution points of the target species to determine the areas suitable for T. radiata. However, several environmental variables, such as elevation, vegetation, human factors, hyperparasitoids, and availability of its host (D. citri) may influence the distribution of the parasitoid but were not considered in the present study. Another important factor to be considered in species distribution modelling is the uncertainties associated with future predictions. Achieving these SRES depends on several factors, like the release of atmospheric greenhouse gases. As a result, these uncertainties should be considered when analyzing the results.
Despite these limitations, our modelling outputs are critical for understanding the factors limiting the distribution of T. radiata for effective biological control programs. In particular, our suitability maps show the importance of using species' physiological stress factors and occurrence records to define species' ecological niches and improve the performance of modelling outcomes. Our suitability maps can be useful for developing biological control programs because the maps can guide ecologists, biologists, plant protection agencies and pest managers to identify suitable areas for mass rearing and releasing the parasitoid.

Conclusion
The potential distribution of T. radiata has been defined globally using the CLIMEX model. The model predicted climate suitable areas outside the present day known distribution regions of the parasitoid. Our model predicted habitat suitability for T. radiata in all continents except Antarctica. The new areas identified as suitable for T. radiata included parts of Europe and Oceania. Habitat suitability for T. radiata will decline from the 2050s to the 2090s under the different climate change scenarios. The distribution maps created using the CLIMEX model may be helpful in the search for and release of T. radiata in new habitats. Moreover, our modeling idea can be adopted by other studies to predict the geographical distribution of biological control agents. www.nature.com/scientificreports/